MediNet

PROJECT NAME

MediNet

DESCRIPTION OF THE PROJECT

MediNet is a privacy‑first platform for collaborative AI on medical imaging. It lets hospitals and research centers improve models together without moving raw patient data by using Calimero for secure coordination and auditability. Only encrypted model updates and minimal metadata flow over a P2P channel; institutional data stays on‑prem at all times. Our demo follows a round‑based loop: local train/infer → publish encrypted updates → receive a stronger global model. 

For cross‑organization evaluation, we’re exploring trusted execution environments (TEE) - e.g., iExec to add runtime confidentiality and access management.

TELEGRAM USER NAME

- @BartoszSol

- @coredumped7893

- @norbertkulus

REPOSITORY WITH THE PROJECT'S CODE

Main repositories:
- https://github.com/EthRome25/cnn-network-poc

- https://github.com/EthRome25/calimero-model-training

Organization: https://github.com/EthRome25

VIDEO DEMO

https://youtu.be/b-F5kKyKNoc

BOUNTIES

Calimero: We have used Calimero for a secure P2P connection between medical entities to exchange data about the current AI model.

Chainbound: We have explored the potential usage of TEE for medical, privacy-focused data with the iExec TEE.

iExec: We had the idea to use iExec TEE to minimize the risk of private data being leaked during execution invocation. The TEE allows for easy access management while at the same time making the model accessible to every trusted entity without the need to have their own infrastructure.